A data-driven review of Tether’s transactions, “users,” and relationships
Update: We have published a new review comparing Tether to other stablecoins.
Update 2: We have released a stablecoin page for comparing the supplies & prices of different USD stablecoins.
Much has been written by passionate critics on Tether, the supposed 1–1 USD-backed token issued by Tether Limited. Some see it as today’s version of Willy, the Mt Gox trading bot that manipulated Bitcoin’s price in 2013/14. Most of the analysis has focused on off-chain dynamics: Tether’s lack of a banking partner, its split from its auditor, the ownership relationship vis-a-vis Bitfinex, etc.
Because Tether is a blockchain token, signals can be derived from the on-chain data as well. My partner Edwin Ong and I have been collecting and analyzing data on different blockchains (including Ethereum usage) for some time, and we recently took a closer look at Tether.
**Tether Blockchain Background**First, a primer on how Tether’s blockchain works: Tether is a token issued on the Omni Layer (“Mastercoin” to old-school blockchain enthusiasts), a platform that enables the creation of new assets on the Bitcoin blockchain.
Every Omni transaction (and thus Tether transaction) is recorded in a Bitcoin transaction sharing the same transaction hash. For example, this BTC transaction is actually the sending of 50,000 USDT .
To date, Tether has “issued” 2.22 billion USDT tokens on the Omni network. (Separately, Tether has also “authorized” 60M USDT ERC20 tokens on the Ethereum network.)
The hockey stick graph of Tether’s token supply over time
While other coins do run on Omni, Tether is the largest tenant by far, with over 75% of Omni’s transactions ever and over 95% of its transactions in January. At this point, Omni is effectively a Tether blockchain.
Tether accounts for the vast majority of Omni’s transactions
Despite its age, the Omni ecosystem is relatively sparse. Many of the repos on Omni team’s Github are very much a work in progress. The primary user-facing tool is https://omniexplorer.info, a block explorer that is today mostly used to track USDT movement, especially by users from Binance and Remitano (a Vietnamese exchange that is now a top website in that country).
OmniExplorer referral traffic. (via SimilarWeb)
Prior to splitting with Tether, the auditor Friedman LLP wrote that it
“utilized the Omniexplorer.info website to trace that the balance per the Omniexplorer.info website matches the information provided by the Client.”
Even the carefully thought out TetherReport relied on collections of “500 sequential transactions pulled from OmniExplorer.”
**Our Approach**While OmniExplorer is useful, there is no substitute for primary data when it comes to blockchain analysis. To collect Omni data, we ran full nodes of Omnicore and synced them to our data toolchain.
**Transaction Trends**We first looked at Tether’s transaction trends. Unsurprisingly, the big picture is what you might expect: both the number of transactions and the aggregate value of transactions have skyrocketed in recent months.
The aggregate value of Tether transactions nearly doubled each of the past 3 months.
To put the transaction data in perspective, we looked at Tether from the MV=PQ lens framed by Chris Burniske:
Average daily supply in 2017 = 273,374,613 (M)Total 2017 transaction value = 14,963,834,982 (PQ)V = PQ / M = 14,963,834,982 / 273,374,613 = 54.73
Tether’s velocity of 54.7 is about 10 times higher than that of the M1 money supply. (Of course, speculation is typically not included in money circulation calculation.) Regardless, users sure love moving Tethers! So, who are they?
Tether Users To answer the user questions, we mapped out different distributions of Tether usage by addresses, and created lists of the most frequent and the biggest Tether senders.
Tether’s biggest s[p]enders
The #1 most frequent user, with its iconic address 1Po1oWkD2LmodfkBYiAktwh76vkF93LKnh, is the exchange Poloniex, sending over 2B USDT (nearly Tether’s entire supply!) over 111K transactions, averaging nearly 19K USDT per transaction. Clearly, with their high number of transactions, it’s obvious that the busiest addresses are exchange wallets.
Tether’s busiest senders
The biggest senders list looks a little different. The #1 sender is 1KYiKJEfdJtap9QX2v9BXJMpz2SfU4pgZw, Bitfinex’s account (per Tether), sending nearly 5B USDT over 38K transactions, averaging nearly ~130K USDT per transaction, a lot higher than Poloniex’s 19K per transaction.
Some of these other big senders are not as busy. 1NTMakcgVwQpMdGxRQnFKyb3G1FAJysSfz is Tether’s new “treasury account” after the November hack, distributing 1.7B of newly printed USDT in 31 transactions, mostly to 1KYiKJEfdJtap9QX2v9BXJMpz2SfU4pgZw (Bitfinex). 1HckjUpRGcrrRAtFaaCAUaGjsPx9oYmLaZ (the Chinese platform Huobi) averages 12.6M USDT per transaction over 62 sends!
Recycle
12.6M is a lot of Tethers per transaction. It turns out that in Huobi’s case, the transactions are all to the same recipient.
Those are some big numbers per transaction!
Our tool lets us easily track the interactions between any 2 addresses, so we can see that Huobi’s address has been sending those tens of millions everyday to 1LAnF8h3qMGx3TSwNUHVneBZUEpwE4gu3D, another Huobi address.
This other Huobi address maintains a nearly 50/50 split in terms of aggregate amount sent and received.
This Huobi address maintains an almost 50/50 sent/received ratio
This 50/50 split is hardly atypical for Tether addresses. There are many more who maintain this ratio, whether they transact frequently (like 16bUGjvunVp7LqygLHrTvHyvbvfeuRCWAh) or not (like 1MMZ5QNzARpn7kvf2aj4UzViALP9j84Ufi).
What this suggests is that many Tether addresses are simply being used to cycle funds from entities to entities. Thus, the high token velocity should be discounted for such use cases.
Address Relationships
We also mapped out relationships between distinct addresses, focusing on relationships between addresses that transact together often and addresses that transact large numbers of USDT together.
Some of these addresses love sending Tethers to their favorite receivers many many times
Bitfinex’s dominance in Tether is absolutely clear from examining these interactions. 1KYiKJEfdJtap9QX2v9BXJMpz2SfU4pgZw alone is involved in 22 of the top 50 highest-valued “relationships.”
One of Bitfinex’s addresses dominates the “Highest Valued Relationships” chart
By mapping out these relationships on Tether’s blockchain and following a few degrees of separation, we have been able to track and in some cases correlate capital flow, especially in combination with other blockchain and market data.
**Balance Distribution**Another area we examined was the distribution of USDT holdings amongst addresses. Tether.to’s top 20 addresses is a popular page, and we are adding a top 100 list to the mix. While blockchain address distribution can hardly be conclusive given the ease of new address generation, the data can still be informative.
In Tether’s case, the top 200 addresses out of Tether’s nearly 100K active addresses hold over 2B USDT. Yes, the top 0.2% owns over 90% of the token’s total supply. This is more than double BTC’s wealth concentration.
**Data Download**We believe that on-chain data, in combination with network metrics, ultimately provides the best understanding of the vibrancy and health of any blockchain. They will matter as much tomorrow as DAUs and ARPUs do today.
Given the dearth of easily downloadable on-chain data for Tether, we are publicly releasing the entirety of Tether’s transaction data up to block 507015.
This dataset is a little bigger than what Google Sheets or Excel can easily handle, so storing the data in a database or data warehouse might make analysis more pleasant. A graph database can be useful if you want to follow the “social graph” of the addresses. If you are just interested in high-level trends, you can also browse the data we have published on Blockspur.
Tether’s blockchain data has answered a few of our questions and raised some new ones, and we are excited to hear if they can provide new insights for you. If you have specific questions about using the data or about this post in general, please contact us and we’d be happy to share our perspectives.
Thanks to Brian Schroeder, Gordon Wilson, and Remy Hathaway for feedback. For additional data-driven blockchain posts, please see our work on Ethereum usage and Zcash.